Machine Learning Overview (page 7 of 13)

Variables are observable and measurable. For instance, an employee’s education level or their gender are variables, which vary for each observation or example. A parameter is a numerical property of the model (not of an example or rwo of data).

A hyper-parameter is a value we (the analyst) set as an input to the machine learning algorithm. Hyper-parameters are conceptually similar to architectural blue prints that guide the construction process. The machine learning algorithm uses those hyper-parameters when it is training the model. We usually set or tune these hyper-parameters through trial and error or more objectively using a Python algorithm.